How-to-Use-Customer-Analytics-to-Create-Personalized-Airline-Offers_Blog.png

With the increased availability of big data, the notion of the customer has become more granular. Air travel marketers no longer segment the market based on gross demographic or purchase characteristics (age, total travel spend).

Instead marketers can segment by 100s of other characteristics or behaviors including social media activity, whether potential customers searched travel options on an OTA or not, and whether they responded to their last promotion.

Unfortunately, traditional statistical analysis is not designed to handle a multitude of relevant variables built around individual traveler behaviors. Statistical work quantifies "average" relationships. It is normally not suited for specification of multiple models that apply - with different variables and different coefficients - to different customers or segments.

Let’s take a highly simplified example. Consider four travelers, all of whom purchased the travel product but who exhibited different behaviors online:

  Searched OTA Respond to E-mail Used TripAdvisor
Traveler #1 No Yes No
Traveler #2 Yes No Yes
Traveler #3 Yes No No
Traveler #4 Yes No No

 

In this limited dataset, "Searched OTA" would be identified as the most important explanatory variable for booking: most travelers who booked had searched OTA. This view however, might well ignore that the e-mail campaign was effective, even if it didn’t drive all of the bookings. Similarly, TripAdvisor would be eliminated as a critical factor - even though it actually could have been important to Traveler #2 in conjunction with searching OTA. "Averages" do a poor job of describing the behavior of the individuals in this group of travelers.

In the real world, it’s of course even more complicated. The limitations of averages are magnified across diverse customers and diverse segments. Correlations across various customer segments often drive misleading conclusions - attributing success to factors that don’t apply well to all segments while eliminating factors that apply only to a subset of individuals.

A "Segment of One" is the ultimate marketing perspective that translates into personalized offers based on individual-specific behavior. Rather than pigeon-hole customers into pre-ordained classifications, marketing designs campaigns around the specific behaviors each traveler has actually exhibited.

In the above example, a supplier could personalize campaigns. Rather than dismiss e-mail and TripAdvisor as important elements in marketing, a supplier could target an email campaign to Traveler #1 and evaluate TripAdvisor with respect to booking success with Traveler #2. Greater granularity is important in:

  • Targeting relevant promotions and campaigns. Each individual traveler responds differently to promotions. Marketing departments need to design promotions around those specific individuals for whom they have proven successful.
  • Evaluation of marketing spend in different channels. Rather than evaluate campaigns on average, resources can be used more efficiently by more targeted campaigns.
  • Continuous refinement of offers and promotions based on individual traveler results. More accurate data on individual behavior will lead to more successful individualized campaigns. Every customer interaction is an opportunity to learn and adjust.

Depending on the situation, "Segment of One" can drive more or fewer interactions, based on individual customer behavior:

  • More Interactions for some. "Segment of One" can offer more effective exploitation of multi-channel marketing. Travel booking continues to be quite complicated on average, with customers using a number of sites over days or weeks before making a booking. However, there is no one process travelers follow. Thus, individualized campaigns recognize the propensity for different travelers to follow different processes, utilizing different channels and different devices, before making a booking.  Individualized marketing translates into the right combination of interactions or campaigns for each individual.
  • Fewer Interactions for others. "Segment of One" will limit promotions to those who are most likely to respond. A corollary of individualized marketing is that some customers will receive fewer offers – customers will not receive offers or be the focus of campaigns that historically have not been effective for them individually.

Approaches are statistically based, around individual activity:

  • Averages or probabilities are applied on an individual basis. Each person’s personal history is analyzed to determine whether he always searches OTA’s or whether he ever responds to an email promotion. No one’s behavior is averaged with other customers; every one’s own (often inconsistent) history is analyzed for tendencies and "averages" (In fact, the most sophisticated marketers will model individual’s different behaviors in different contexts - yet another level of granularity.)
    • Conversion rates of 5% - tremendous for many broad promotional initiatives - can instead reach over 70% when promotions are based on historic individual behaviorTraditional segmentation (demographics, etc.) becomes a supplement to the "Segment of One".
  • Traditional segmentation may offer the opportunity to test the impact of campaigns for which there is not enough data in the individualized database.

As big data stores interactions across all channels and all devices, increasingly, it makes more sense to design campaigns around each individual’s actual historic behavior:  "Segment of One".


Read about the potential of customer data for airline merchandisers to deliver personalized experiences.

Read the blog